Explaining and Sustaining AI Assimilation in Higher Education: The Dual Roles of Engagement and Skeptical Engagement

Although artificial intelligence (AI) is increasingly embedded in higher education, its assimilation into academic writing remains uneven and theoretically underexplored. Drawing on self-determination theory (SDT) and the perceived risk facets model, this study tests an integrated framework of how motivational enablers and risk-based inhibitors jointly shape AI assimilation in formative writing. A sequential multi-method design was used: qualitative analysis of reflective essays from 22 students identified enablers and inhibitors, which informed a survey of 303 students, analysed using partial least squares structural equation modelling (PLS-SEM). Autonomy support, AI empathy, explainability, and perceived fairness foster engagement, whereas technostress, ethical concerns, and loss of control cultivate skeptical engagement, a cautious yet constructive mode of use. Both engagement and skeptical engagement positively predict assimilation, indicating that risk awareness does not suppress adoption but promotes reflective integration. By reconceptualising resistance as skeptical engagement, the study offers a dual-pathway framework explaining how motivation and critical thinking jointly drive AI integration into routine practice, extending motivation and risk theories to AI-mediated learning and yielding implications for autonomy-supportive pedagogy, transparent AI design, and governance that legitimises critical rather than uncritical use.